JP2001216092A - Data writing method, data reading method, disk drive and disk drive controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the reading efficiency to a write command that is kept waiting for its execution and also to improve performance of a hard disk drive. SOLUTION: When a data read request is received while a data write request is kept waiting for its execution, the logical block address of a write command W1 is compared with that of a read command R for deciding presence or absence of their duplication in a positional relation deciding step and a data reading step, where the read command R is executed before all write commands W are executed when no duplication of block addresses is decided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to reading and writing data in a disk drive.
In particular, the present invention relates to a method for reading and writing data quickly.

[0002]

2. Description of the Related Art When a host computer issues a plurality of write requests to a hard disk drive as an external storage device of a computer, it takes a considerable amount of time to actually write data to the hard disk drive and finish the operation. Time is needed. Therefore, the hard disk drive notifies the host computer of the reception of the command related to the write request, while holding the write command waiting to be executed.

When there are a plurality of write commands waiting to be executed, an RPO (Rotational Position Optimizati
On), commands to be sequentially executed are selected. When the RPO executes a command waiting to be executed, the seek time from the start of a seek to a target track on the magnetic disk to the arrival at the track, and the start of access to the target sector after the arrival. Predict the disk rotation latency. In this method, the command waiting for execution with the shortest sum of the seek time and the rotation waiting time is selected as the next command to be executed.

[0004]

Since the execution time of a command can be shortened by the RPO, the performance of the hard disk drive is improved. But,
It is needless to say that it is desirable to further shorten the command execution time by RPO to improve the performance of the hard disk drive. By the way, in the hard disk drive, there may be a case where a request to read data stored in the magnetic disk is made from the host computer in a state where there is a write command waiting to be executed. In such a case, in the conventional hard disk drive, all the write commands waiting to be executed are executed, that is, the data is read by executing the read command after the writing to the magnetic disk is completed.
Therefore, when there is a write command waiting to be executed, it takes a considerable time before reading is performed. Also from this point, it has been desired to improve the performance of the hard disk drive. Accordingly, an object of the present invention is to improve the performance of a hard disk drive by improving the efficiency of writing or reading when there is a write command waiting to be executed.

[0005]

SUMMARY OF THE INVENTION The present invention relates to a method of writing data to a recording medium for writing write data for which a write request has been made by a host device, the logical block address of a preceding write command in a write execution waiting state. And a logical block address of a subsequent write command for which a new write request has been made, and a positional relationship determining step of determining a positional relationship on the logical block address between the preceding write command and the subsequent write command; When it is determined by the positional relationship determination step that the preceding write command and the subsequent write command are separated on a logical block address, a separation that determines whether the separation distance is equal to or less than a specified separation distance. The distance determination step and the separation distance determination step When it is determined that the separation distance on the logical block address between the preceding write command and the subsequent write command is equal to or less than a prescribed distance, the write data corresponding to the preceding write command and the write corresponding to the subsequent write command are written. The above object is solved by a data writing method, comprising: a data writing step of writing data continuously to the recording medium.

In the data writing method according to the present invention, when there are a plurality of preceding write commands for which the separation distance is determined to be equal to or less than a specified separation distance in the separation distance determination step, the data writing step includes the following. Immediately after writing the write data corresponding to the preceding write command, it is desirable to write the write data corresponding to the subsequent write command having the smallest separation distance.

According to the present invention, there is provided a data reading method when a data read request is issued from a host device in a state where a data write request is waiting for execution. Comparing the logical command with the logical block address of the read command for which the read request has been made, and determining whether the write command and the read command have an overlap on the logical block address. A data reading step of executing the read command before all of the write commands are executed, when it is determined by the relationship determining step that the write command and the read command have no overlap on a logical block address; Characterized by having The data reading method solving the problem.

In the data reading method of the present invention, the execution of the read command is performed by executing the write command in which a distance between a leading end of a logical block address of the read command and a trailing end of the logical block address is minimized. It can be performed subsequently.

The present invention also provides a disk-shaped recording medium which can be accessed randomly, a command holding means for holding a write command transmitted from an external device until the command is executed,
Command relation determining means for determining whether there is an overlapping portion of a logical block address between a write command held in the command holding means and a write command newly transmitted from the external device; and When the means determines that there is no overlapping portion, whether the write command held by the command holding means and the newly transmitted write command are separated from a specified distance on a logical block address Command distance determining means for determining whether the write command held by the command holding means and the newly transmitted write command are not separated from the specified distance by the command distance determining means. Then, the write data corresponding to the two commands is continuously written to the storage medium. Is a disk drive apparatus characterized by comprising a writing instruction means for instructing to write to.

In the disk drive of the present invention,
The command distance determination unit is configured such that, when there are a plurality of write commands held in the command holding unit in which the distance is equal to or less than a specified value, the distance becomes minimum with respect to the newly transmitted write command. The write command held by the command holding means can be specified.

Further, the present invention provides a disk-shaped recording medium which can be accessed randomly, a command holding means for holding a write command transmitted from an external device until the command is executed,
A command comparing unit that determines a relationship between a read command for a read request transmitted from the external device and a write command held in the command holding unit in a logical block address, based on a comparison result of the command comparing unit And a read instructing means for instructing reading.

In the above disk drive device, the command comparing means can determine whether or not the read command overlapping the write command on a logical block address exists. Also,
In the above disk drive device, the read instructing unit is held in the command holding unit when the command comparing unit determines that the read command overlapping on the logical block address with respect to the write command exists. After executing the duplicated write command among the write commands, the command holding unit can be instructed to execute the read command. Further, the read instructing means, when the command comparing means determines that the write command and the read command do not overlap on a logical block address, holds the write command awaiting execution in the command holding means. To execute the read command.

Further, according to the present invention, there is provided a controller of a disk drive for controlling writing of data to a disk-shaped recording medium which can be accessed randomly, and holds the data until a preceding write command previously transmitted from an external device is executed. Command holding means, and a command comparison means for determining whether or not there is the preceding write command that is separated within a range equal to or less than a specified value on a logical block address and a subsequent write command newly transmitted from the external device, When the command comparing means determines that there is the preceding write command that is separated within a range equal to or less than the specified value, a write instruction for instructing that the subsequent write command be executed with priority over the preceding write command. Disk drive device comprising: Controller is provided for. Further, according to the present invention, a controller of a disk drive device for controlling reading of data from a randomly accessible disk-shaped recording medium, which holds a plurality of pre-write commands transmitted from an external device in advance. Command holding means, command comparison means for determining whether or not there is a preceding write command duplicated on a logical block address with a read command newly transmitted from the external device, and the command comparison means Read instruction means for instructing execution of the read command prior to execution of the write command when it is determined that there is no duplicate preceding write command on a logical block address. Disk drive device control There is provided.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described based on embodiments. FIG. 1 shows a hard disk drive (HDD) device 1 as an external storage device of a computer for implementing the present invention.
FIG. 2 is a diagram showing a schematic configuration of a zero. As shown in the figure,
The HDD device 10 includes a magnetic disk 21 that stores data as a storage medium, and a spindle motor 22 that drives the magnetic disk 21 to rotate. The magnetic head 23 performs recording / reproducing (writing / reading) of data to / from the magnetic disk 21. The head arm 24 has a magnetic head 23 at its tip, and moves over the recording surface of the magnetic disk 21. The actuator 25 holds the head arm 24 and drives the head arm 24 to rotate. As a result, the magnetic head 23 moves in a substantially radial direction of the magnetic disk 21 and
Is configured to be able to randomly access the recording surface.

Magnetic disk 21, spindle motor 2
2. The drive mechanism including the magnetic head 23 and the actuator 25 is controlled by the control circuit 30. The control circuit 30 has an HDC (Hard Disk Control
r) 31, ROM 32, RAM 33 and host I / F
(Interface) 34, which are interconnected via a bus 35.

The HDC 31, as an internal controller of the HDD device 10, controls the entire HDD device 10 according to a control program and control data stored in the ROM 32. The HDC 31 executes arithmetic processing for servo control and control when writing and reading data. Thus, the spindle motor 22 and the actuator 25 are driven, and writing / reading is performed using the recording head and the reproducing head of the magnetic head 23. The HDC 31 stores a part of the data recorded on the magnetic disk 21 as a storage medium in the RAM 33, and reads a part of the data stored in the magnetic disk 21 in advance and stores the data in the RAM 33. Control.

The ROM 32 stores a control program executed by the HDC 31 and control data used by the control program. RAM 33 is
It has a function as a cache memory for temporarily storing write data recorded on the magnetic disk 21 and for temporarily storing read data read from the magnetic disk 21. Also, the host I / F 34
Is an interface circuit for transmitting and receiving data and commands to and from the host 40.

FIG. 2 shows main functional blocks of the HDC 31. The HDC 31 includes an MPU (micro processing unit) 41, an MM (memory manager) 42, a RAM 43, and an SG 44 (segment handler). The RAM 43 stores a table having the configuration shown in FIG. 16 and a write segment having the configuration shown in FIG. 17 in a table format. Note that FIG.
The table of “NO OF HIT (number of overlay hits)”, “HEAD BEST SKIP LENGTH”, “HEAD BEST SKIP LENGTH” and “HEAD BEST SKIP S
EGMENT ID), Tail best skip length (T
AIL BEST SKIP LENGTH) and tail best skip segment ID (TAIL BEST SKIP SEGMENT ID) are stored. The write segment in FIG. 17 is composed of four words, one word and two words out of four words.
The LBA uses the word, the block length (LENGTH) uses the third word, and the hit status (HIT STATUS) uses the fourth word. SG44
Accesses the RAM 43 through the MM 42 and
A search of the table in 43 is performed. The SG 44 has a register for temporarily storing a skip length (SKIP LENGTH) described later.

In the present embodiment, the positional relationship between a preceding write command already transmitted from the host 40 and a subsequent write command newly transmitted from the host 40 on a logical block address (hereinafter, LBA) is determined. .
The LBA information, that is, the leading and trailing LBAs and block lengths of the write command are included in the host 4 data along with the write command and write data.
Sent from 0. This write data is stored in RAM3
3 and information on the LBA is stored in the write segment shown in FIG. In the present embodiment, the preceding write command is referred to as buffer data, and the subsequent write command is simply referred to as a write command. In the present embodiment, L
The positional relationship on the BA may be simply referred to as a relationship or a positional relationship.

In the present embodiment, as a first stage, the relationship between the buffer data and the write command is represented by N shown in FIG.
O HIT-A, TAIL_SEQ, OV_TAIL, O
V_ALL, OV_COVER, OV_HEAD, HEA
It is determined based on the LBA which of the eight classifications D_SEQ and NO HIT-B falls. This determination is made based on the flows shown in FIGS. Next, as a second stage, LBA is also used and FIGS.
The relationship between the buffer data and the write command is determined in detail based on the flow shown in FIG. The HPO according to the present embodiment executes the RPO based on the determination result.
It is a DD device.

First, judgment is made in the first stage based on FIG.
The two categories will be described. In FIG. 3, the LBA at the end of the buffer data write destination is denoted as LLBA. The block length of the buffer data is represented by LENGTH. Further, the leading LBA at the write destination of the write command is referred to as SLBA, and the LBA at the trailing end is referred to as ELBA. Furthermore, the block length of the LBA of the write command is referred to as SLEN. Items denoted by # 1 to # 8 in FIG. 3 indicate write commands. Note that #
1 to # 8 are written for the purpose of classifying the write commands for the sake of convenience, and it is necessary to first deny that they do not indicate the order of transmission from the host 40.

In the present embodiment, the positional relationship of the write command with respect to the buffer data is classified into eight types as shown in FIG. The contents of each are as follows. <NO HIT-A> The write command # 1 exists at a position separated from the LLBA of the buffer data. Thus, the write command # 1 for the buffer data
Has no overlap, NO HIT-A
Judge.

<TAIL_SEQ> Write command #
When the leading end of the write command # 2 is continuously connected to the trailing end of the buffer data as in
Judge as L_SEQ. <OV_TAIL> When the buffer data and the write command # 3 overlap at the rear end of the buffer data as in the case of the write command # 3, it is determined to be OV_TAIL. <OV_ALL> Write command # 4 is all contained in the buffer data. In such a case, O
Judge as V_ALL.

<OV_COVER> Write command #
5 includes all buffer data. In such a case, it is determined as OV_COVER. <OV_HEAD> When the buffer data and the write command # 6 overlap at the leading end of the buffer data as in the case of the write command # 6, it is determined to be OV ## HEAD.

<HEAD_SEQ> Write command #
When the rear end of the write command # 7 is continuously connected to the front end of the buffer data, as shown in FIG.
Judge as D_SEQ. <NO HIT-B> Like write command # 8,
Write command # 8 at the leading end of buffer data
NO HIT if is far from the buffer data
Judge as -B.

The above determination is made by the SG (segment handler) 44 of the HDC 31 comparing the LBA information of the write command with the LBA information of the segment table stored in the RAM 33. A specific flow will be described with reference to FIGS. 4 to 6, in step S11, LLBA <SL
It is determined whether BA-1 or not, and LLBA <SLBA-1
If so, the process proceeds to step S12; otherwise, the process proceeds to step S13. The determination in step S11 is a comparison of the positions of the LBA at the rear end of the buffer data and the LBA at the front end of the write command. LL
If BA <SLBA-1, in step S12, LBA is inserted between the buffer data and the write command.
Since there is no overlap, the relationship of the write command to the buffer data, that is, the hit status (HIT
STATUS) is determined to be NO HIT-A. Note that "A"
Indicates that the write command exists after the buffer data.

In step S13, LLBA = SLBA-
1 or LBA at the end of buffer data
It is determined whether or not the LBA at the leading end of the write command is continuous. If they are consecutive, the hit status is determined to be TAIL ## SEQ in step S14. Unless LLBA = SLBA-1, step S15
(Hereinafter, FIG. 5).

In step S15, LENGTH <LL
It is determined whether or not BA-SLBA + 1. If so, go to step S19; otherwise, go to step S1.
Proceed to 6. In step S16, it is determined whether or not LLBA <ELBA. If LLBA <ELBA, the hit status is set to OV_TAI in step S17.
L, otherwise, the hit status is determined to be OV_ALL in step S18. On the other hand, in step S19, it is determined whether or not LLBA ≦ ELBA, and if so, in step S20, the hit status is determined to be OV_COVER. Otherwise, the process proceeds to step S21.

In step S21, LENGTH <LL
It is determined whether or not BA-ELBA, and if so, the hit status is set to NO HIT in step S22.
Judge as -B. Otherwise, the process proceeds to step S23 (hereinafter, FIG. 6). In step S23, LENGTH
= LLBA-ELBA, and if so, the hit status is determined to be HEAD_SEQ.
On the other hand, if not, the hit status is set to OV_HE.
Judge as AD. This is the end of the first-stage determination.

Next, the processing of the second stage will be described with reference to FIGS. <NO HIT-A> If NO HIT-A is determined, the determination shown in FIG. 7 is made. That is, in step S101, skip length (SKIP LE)
NGTH) Skip Criteria (SKIP CR)
It is determined whether or not the value is larger than (TERIA). Here, the skip length is a value defined by LLBA-SLBA + 1, and indicates the distance between the rear end of the buffer data and the front end of the write command. If the skip length is larger than the skip criteria, it is determined in step S102 that the write command is NO HIT for the buffer data. In other words, if the distance between the leading end of the write command and the trailing end of the buffer data exceeds the skip criterion, which is the prescribed distance, it is determined to be NO HIT. If NO HIT is determined, it is considered that there is no relationship between the buffer data and the write command, and no processing such as continuous writing is performed when data is written.

If the skip length is equal to or less than the skip criterion, the write command
At 103, it is determined that the data is a tail hit (TAIL HIT) for the buffer data. In other words, although the buffer data and the write command are separated from each other, since the distance is equal to or less than a specified value, the buffer data and the write command are regarded as data related to continuity. And Therefore, it is effective for performance improvement by RPO. The same applies to the case where NO HIT-B is determined. If it is determined that the tail hit has occurred, the tail best skip length (T
It is determined in step S104 whether or not (AIL BEST SKIP LENGTH) is less than 0. Tail
If the best skip length is less than 0, the process ends. Otherwise, in step S105,
It is determined whether or not the current skip length is smaller than the tail best skip length. If it is smaller, in step S106, the current skip length is set as the tail best skip length, and the tail best skip length and the tail best skip segment ID are updated. Otherwise, the process ends.

Here, the tail best skip length will be described with reference to FIG. FIG.
Assume that the write command is a tail hit for any of the two buffer data. However, it can be seen that the skip length of buffer data with a segment ID of 2 is smaller than that of buffer data with a segment ID of 1. In this case, the tail best skip length of the write command becomes the skip length with the buffer data having a segment ID of 2, and the tail skip segment length
The ID becomes 2. And the tail best skip
Length and tail skip segment / ID
Are stored in the table shown in FIG.

<TAIL_SEQ> When it is determined that TAIL_SEQ is present, the processing shown in FIG. 8 is performed. TAI
In the case of L_SEQ, the hit status is determined to be a tail hit in step S201. Next, it is determined in step S202 whether or not the current tail best skip length is less than 0. Tail
If the best skip length is less than 0, the process ends. Otherwise, in step S203,
It is determined whether or not the skip length is smaller than the tail best skip length. If it is smaller, in step S204, the tail, the best, the skip,
Update the length and tail best skip segment ID.

<OV_TAIL> When it is determined that OV_TAIL, the processing shown in FIG. 9 is performed. First, in step S301, the hit status is determined to be a tail hit, and then, in step S302, it is determined whether the skip length is smaller than the tail best skip length. If smaller, step S30
At 3, update the tail best skip length and the tail best skip segment ID. If not, the process ends.

<OV_ALL> If it is determined that OV_TALL, the processing shown in FIGS. 10 and 11 is performed. First, in step S401, LLBA and ELB
It is determined whether or not A matches. If they match, the process proceeds to step S402. If they do not match, the process proceeds to step S408 in FIG. In step S402, SLBA
= LLBA-LENGTH + 1 is determined. If so, the process proceeds to step S403; otherwise, the process proceeds to step S405. In step S404, the SLBA and the leading LBA of the buffer data in step S402
If the hit status is determined to match, the hit status is determined to be cover overlay, and then step S40
At 4, increase the overlay hit count by one. In step S405, the hit status is determined to be a tail hit. Next, in step S406, it is determined whether the skip length is smaller than the tail best skip length. If smaller, in step S407, the tail best skip length and the tail best skip segment ID are updated. If not, the process ends.

In step S408, SLBA = LLBA
It is determined whether or not -LENGTH + 1. If so, go to step S409; otherwise, go to step S41.
Proceed to 2. In step S409, the hit status is determined as a head hit (HEAD HIT). Subsequently, in step S410, the skip length (here, SLEN) is changed to the head best skip.
Length (HEAD BEST SKIP LENGT
H) It is determined whether or not it is smaller. If so, the process proceeds to step S411; otherwise, the process ends. In step S411, the head best skip
Length and Head Best Skip Segment ID (HEAD BEST SKIP SEGMENT
ID) is updated, and the process is thereafter terminated.

In S412, the hit status is determined to be all overlay (ALL OVERLAY), and then in step S413, the count of overlay hits is incremented by one.
Increase the number and end the process. Note that the count of the overlay hit is determined by the NO OFH in the table shown in FIG.
Increased in IT cells.

<OV_COVER> When it is determined that OV_COVER is satisfied, step S50 is executed as shown in FIG.
In step S502, it is determined that the hit status is all overlay.
The hit count is incremented by one, and the process ends.

<OV_HEAD> When it is determined that OV_HEAD, the processing shown in FIG. 13 is performed. OV_H
In the case of EAD, the hit status is determined to be a head hit in step S601. Then, step S
At 602, it is determined whether the skip length (in this determination, (LLBA-LENGTH + 1) -ELBA) is smaller than the head best skip length. If smaller, in step S603,
Update the head best skip length and head best skip segment ID. Otherwise, the process ends.

<HEAD_SEQ> If HEAD_SEQ is determined, the processing shown in FIG. 14 is performed. HE
In the case of AD_SEQ, the hit status is determined to be a head hit in step S701. Next, in step S702, it is determined whether or not the best hit skip length so far is negative. If the result is negative, the process ends. If not negative, step S703
Proceed to. In step S703, it is determined whether the skip length is smaller than the head best skip length. Note that the skip length here is zero. If the skip length is smaller than the head best skip length, in step S704 the head best skip length and the head best skip segment ID are updated, and the process ends.

<NO HIT-B> If NO HIT-B is determined, the process shown in FIG. 15 is performed. That is, in step S801, (LLBA-LEN
GTH + 1) -ELBA is determined to be greater than the skip criterion. If the skip length is greater than the skip criterion, step S802
In, the hit status is changed to no hit (NO
HIT). Skip length is skipped
If the condition is equal to or less than the criteria, it is determined in step S803 that the hit status is a head hit, and the process proceeds to step S804. In step S804, it is determined whether the skip length is smaller than the head best skip length. Here, the skip length is (LLBA-LENGTH + 1) -EL
Defined by BA. If the head best skip length is less than or equal to the skip length, the process ends. If the skip length is smaller than the head best skip length, in step S805, the head best skip length and the head best skip segment ID are updated, and the process ends.

As is apparent from the above description, in the present embodiment, the relationship between the buffer data and the write command is first roughly classified as shown in FIG. 3, and then shown in FIG. 7 to FIG. As described above, the technique of judging the relationship between the buffer data and the command in detail is adopted. Note that NO HIT-A, TAIL ## SEQ, OV_TA
IL, OV_ALL, OV_COVER, OV_HEA
D, HEAD_SEQ and NO HIT-B and FIGS.
FIG. 24 shows the relationship with each hit status determined in FIG. 15, including the positional relationship with the buffer data.

The above determination is executed by the segment handler (SG) accessing the RAM 43 storing the segment table through the memory manager (MM) every time a write request is made from the host 40. The write segment constituting the segment table is composed of four words as shown in FIG. The LBA uses the first and second words, the third word uses the block length (LENGTH), and the fourth word uses the hit status. In the conventional HDD, only one set of the write segment exists, and the fourth word is not used. In contrast, in the present embodiment, FIG.
64 sets of write segments shown in FIG. 6 are prepared, and the fourth word which has not been used in the past is used for hit status. Therefore, the write command is held in the RAM 43.

The hit status stored in the write segment is represented by RPO (Rotational Position Optimi).
zation). The RPO first executes a command waiting to be executed, and starts a seek to a target track on the magnetic disk 21 until the track reaches the target track, and accesses a target sector after the seek. Predict the disk rotation waiting time before starting. In this method, the command waiting for execution with the shortest sum of the seek time and the rotation waiting time is selected as the next command to be executed.

In this embodiment, by referring to the hit status stored in the write segment as described above, a write command that can be regarded as a series of data among a plurality of write commands can be associated. That is, in cases other than NO-HIT, it is possible to continuously execute writing as a series of commands. In this embodiment, since 64 sets of write segments are prepared as described above,
Among them, there may be a plurality of command blocks as described above. If there are multiple command blocks like this,
According to the conventional RPO, writing is executed first from the command block in which the total of the seek time and the rotation waiting time is the shortest. Of course, there may be a write command that does not constitute a command block, and therefore, the conventional R
The execution order of the write commands is determined according to the PO.

The concept of the above processing will be described with reference to FIG. FIG. 18A shows a plurality of commands to which a write request has been issued from the host 40. The numbers assigned to the respective commands indicate the order of write requests transmitted from the host, and the horizontal axis indicates LBA. In FIG. 18A, based on the above-described comparison of LBAs, it is determined that each of the commands constitutes a series of data groups. However, the commands have no relation to other commands and are determined to be no hits. Therefore, the command ~ is as shown in FIG.
Can be replaced with six data groups A to D as shown in FIG. The order of writing is not the order transmitted from the host 40 but A, B, C, and D.
Rearrange in order.

Next, a description will be given, with reference to FIG. 19, of a process for executing writing of a data group having an overlay hit. As shown in FIG. 19, the LBA of the command (buffer data) of the preceding write request is 1, 2, and 3, and the LBA of the command (write command) of the subsequent write request is 0 and 1. The hit status of the write command for the buffer data is an overlay hit, and blocks of LBA1 are duplicated. If the writing of the buffer data is executed after the writing of the write command is executed simply according to the RPO, the data of the buffer data is overwritten on the data of the write command for the block of LBA1. Since the data of the buffer data is older than the data of the write command, the overwriting causes old data to be written over new data.

Therefore, in the present embodiment, as shown on the right side of FIG. 19, duplicate blocks in the buffer data are erased (DISCARD). Thereafter, writing of buffer data is executed, and then writing of a write command is executed. By doing so, new data is also written for the overlapping part.

Next, a description will be given of a process performed when a host issues a read request command while a plurality of write commands are waiting to be executed. In a conventional HDD device, when a read request is issued while a write command is waiting to be executed, the read request command is executed after all the write commands waiting to be executed have been executed. However, if a read request command is executed after execution of all write commands is completed, data reading is delayed. What should be prioritized now is the read command. Therefore, in the present embodiment, when a plurality of write commands are waiting to be executed and a read request command is issued from the host 40, the read command can be interrupted before all the write commands are executed. did.

When a read request is issued while a plurality of write commands are waiting to be executed, and when the read request command is interrupted in the middle of the write command, it is necessary to pay attention to the read request command and the write request command waiting to be executed. And when there is an overlap. This will be described with reference to FIG. FIG. 20 shows the relationship between one write command and one read command to facilitate understanding. FIG. 20A shows a case where the preceding write command (W in the figure) and the subsequent read command (R in the figure) have no overlap on the LBA. In such a case, there is no problem if the read command R is preferentially executed before the write command W is executed. However, as shown in FIG. 20B, when the write command W and the read command R have an overlapping portion on the LBA, the read command R is preferentially executed before the write command W is executed. There are the following problems.

The read request from the host 40 is to be made for the latest data. FIG. 20 (b)
In the case shown in (1), the data by the write command W is the latest data. As described above, the write command W and the read command R have overlapping LBAs. Regarding this overlapping portion, the data targeted by the read command R is data by the write command W. Therefore, if the read command R is executed before the write command W is executed and the data is actually written, the data read by the execution of the read command R includes a part that is not the latest. This does not mean that the read command R has been properly executed.

Therefore, in this embodiment, when a read request is issued when there are a plurality of write commands waiting to be executed, the position of the LBA of the write command waiting to be executed and the position of the LBA of the read command requested by the host 40 are determined. We decided to confirm the relationship. When there is a write command waiting to be executed having a relationship as shown in FIG. 20B, the read command is interrupted after executing the write command, and the write command is executed prior to other write commands waiting to be executed. I decided that.

FIG. 21 shows an example. In FIG. 21A, a write command waiting for execution of W1 to W4, R
Are read commands requested from the host 40 when the write commands W1 to W4 are waiting to be executed. FIG.
As shown in (a), the read command R and the write commands W2 and W3 have an overlap on the LBA. Therefore, in order to execute the read command R, it is necessary to execute the read command R shown in FIG.
As shown in the figure, write commands W1, W2 and W3
After that, the read command R is executed. After the execution of the read command R, the write command W
4 was decided to be executed. By doing this,
The data read by execution of the read command R includes the latest data. Note that 〜 in FIG. 21B indicates the order of execution.

FIG. 21 shows an example in which the read command R and the write commands W2 and W3 have an overlap in the LBA. However, as shown in FIG. 22A, the read command R and the write command have an overlap in the LBA. When there is no read command R, as shown in FIG.
Can be executed by interrupting before the write command W1.

[0055]

As described above, according to the present invention, the efficiency of writing or reading when there is a write command waiting to be executed can be improved, and the performance of the hard disk drive can be improved.

[Brief description of the drawings]

FIG. 1 shows an HDD (Hard Disk Dr.) according to the present embodiment.
ive) It is a figure which shows the schematic structure of a device.

FIG. 2 shows an HDC (Hard Disk) of the HDD according to the embodiment.
3 is a diagram illustrating a main configuration of a Disk Controller. FIG.

FIG. 3 is a diagram for explaining a relationship between buffer data and a write command.

FIG. 4 is a flowchart for determining the relationship between buffer data and a write command.

FIG. 5 is a flowchart for determining the relationship between buffer data and a write command.

FIG. 6 is a flowchart for determining the relationship between buffer data and a write command.

FIG. 7 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is NO HIT-A.

FIG. 8 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is TAIL_SEQ.

FIG. 9 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is OV_TAIL.

FIG. 10 is a flowchart showing a subsequent process when the relationship between buffer data and a write command is OV_ALL.

FIG. 11 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is OV_ALL.

FIG. 12 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is OV_COVER.

FIG. 13 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is OV_HEAD.

FIG. 14 is a flowchart showing the subsequent processing when the relationship between buffer data and a write command is HEAD_SEQ.

FIG. 15 is a flowchart illustrating a subsequent process when the relationship between buffer data and a write command is NO HIT-B.

FIG. 16 is a diagram showing a table stored in a RAM 43.

FIG. 17 is a diagram showing a configuration of a write segment.

FIG. 18 is a diagram for explaining data writing according to the present embodiment.

FIG. 19 is a diagram for describing data writing at the time of an overlay hit according to the present embodiment.

FIG. 20 is a diagram showing a relationship between a write command and a read command.

FIG. 21 is a diagram illustrating a state when a read command is issued while a plurality of write commands are waiting to be executed.

FIG. 22 is a diagram illustrating a state when a read command is issued while a plurality of write commands are waiting to be executed.

FIG. 23 is a diagram for describing a best tail skip length.

FIG. 24 is a table showing each hit / judgment determined based on FIGS.
FIG. 4 is a diagram showing a relationship with a status including a positional relationship with buffer data.

[Description of Signs] 10 HDD (Hard Disk Drive) device, 21 magnetic disk, 22 spindle motor, 23 magnetic head,
Reference numeral 24: head arm, 25: actuator, 30: control circuit, 31: HDC (Hard Disk Controller), 32:
ROM, 33 RAM, 34 host I / F (interface), 35 bus, 40 host, 41 MPU
(Micro processing unit), 42 ... MM
(Memory manager), 43 ... RAM, 44 ... SG
(Segment handler)

Continuing from the front page (72) Inventor Kuroda Hisashi 1 Kirihara-cho, Fujisawa-shi, Kanagawa Prefecture IBM Japan, Ltd. Fujisawa Office (72) Inventor Jun Kanamaru 1 Kirihara-cho, Fujisawa-shi, Kanagawa Japan M Co., Ltd. Fujisawa Office (72) Inventor Hiromi Nishinomiya 1 Kirihara-cho, Fujisawa City, Kanagawa Prefecture IBM Japan Co., Ltd. Fujisawa Office (72) Inventor Takahiro Saito 1 Kirihara-cho, Fujisawa City, Kanagawa Prefecture BM Corporation Fujisawa Office F-term (reference) 5B014 EB05 5B065 BA01 CA11 CA15

Claims (12)

[Claims] 1. A method for writing data to be written to a recording medium, wherein write data for which a write request has been issued from a host device is provided, wherein a logical block address of a preceding write command in a write execution waiting state and a new write request are issued. A positional relationship determining step of comparing a logical block address of a subsequent write command to determine a positional relationship on the logical block address between the preceding write command and the subsequent write command; When it is determined that the command and the subsequent write command are separated on the logical block address, a separation distance determining step of determining whether the separation distance is equal to or less than a specified separation distance; and By the determining step, the preceding write command and When it is determined that the separation distance on the logical block address from the subsequent write command is equal to or less than a specified distance, the write data corresponding to the preceding write command and the write data corresponding to the subsequent write command are continuously generated. And writing the data to the recording medium. 2. When there are a plurality of preceding write commands for which the separation distance is determined to be equal to or less than a prescribed separation distance in the separation distance determination step, the data writing step includes a step of writing corresponding to the preceding write command. 2. The data writing method according to claim 1, wherein immediately after writing the data, write data corresponding to the subsequent write command having the smallest separation distance is written. 3. A data reading method when a data read request is issued from a host device in a state where a data write request is waiting to be executed, wherein a logical block address of a write command in a write execution waiting state and a new read request are provided. Is compared with the logical block address of the read command
A positional relationship determining step of determining whether there is a logical block address overlap between the write command and the read command; and a logical block address overlap between the write command and the read command by the positional relationship determining step. A data reading step of executing the read command before all of the write commands are executed when it is determined that the data does not exist. 4. The execution of the read command is performed after the execution of the write command in which the distance between the leading end of the logical block address of the read command and the trailing end of the logical block address is minimized. 4. The data reading method according to claim 3, wherein: 5. A disk-shaped recording medium that can be accessed randomly, a command holding unit that holds a write command transmitted from an external device until the command is executed, a write command held by the command holding unit, and a new command from the external device. Command relationship determining means for determining whether there is an overlapping portion of a logical block address between the command and the write command transmitted to the command. Command distance determining means for determining whether the write command held in the holding means and the newly transmitted write command are more than a specified distance on a logical block address, and the command distance determining means The write command held in the command holding means and the new transmission Write instruction means for instructing, when it is determined that the written command is not apart from the prescribed distance, writing data corresponding to the two commands to the storage medium continuously. A disk drive device characterized by the above-mentioned. 6. The command distance judging means, when there are a plurality of write commands held in the command holding means in which the distance is equal to or less than a specified value with respect to the newly transmitted write command, 6. The disk drive device according to claim 5, wherein a write command held by the command holding unit having a minimum distance is specified. 7. A disk-shaped recording medium that can be accessed randomly, command holding means for holding a write command transmitted from an external device until execution, a read command for a read request transmitted from the external device, and the command Command comparison means for judging the relationship in the logical block address with the write command held in the holding means, and read instructing means for instructing reading based on the comparison result of the command comparison means. Disk drive device. 8. The disk drive according to claim 7, wherein the command comparing unit determines whether the read command overlapping the write command on a logical block address exists. apparatus. 9. The read instructing unit is held in the command holding unit when the command comparing unit determines that the read command overlapping on the logical block address with respect to the write command is present. 8. The disk drive device according to claim 7, wherein after executing the duplicate read command among the write commands, the command holding unit is instructed to execute the read command. 10. The read instructing means, when the command comparing means determines that the write command and the read command do not overlap on a logical block address, holds the write command awaiting execution in the command holding means. 8. The disk drive device according to claim 7, wherein an instruction is given to execute the read command in a state where the read command is executed. 11. A controller of a disk drive device for controlling writing of data to a randomly accessible disk-shaped recording medium, the command being retained until a preceding write command previously transmitted from an external device is executed. A command comparing means for judging whether or not there is a preceding write command which is separated from a subsequent write command newly transmitted from the external device in a range not more than a specified value on a logical block address, and the command comparison Means for instructing the subsequent write command to execute writing with priority over the preceding write command when it is determined by the means that there is the preceding write command that is separated within the range of the specified value or less. Control of disk drive device characterized by comprising Troller. 12. A controller of a disk drive device for controlling reading of data from a randomly accessible disk-shaped recording medium, wherein the controller holds a plurality of pre-write commands transmitted from an external device in advance. A command holding unit, a command comparison unit that determines whether there is the preceding write command overlapping on a logical block address with a read command newly transmitted from the external device, and the read command by the command comparison unit. Read instruction means for instructing execution of the read command prior to execution of the write command when it is determined that there is no duplicate preceding write command on a logical block address. Disk drive controller .